Neural Network Predictions of the Impact Behaviour of GFRP Laminates

Composite materials are widely used in aerospace, naval and automotive applications due to their high strength and stiffness to weight ratio. Nevertheless, damages related to low-velocity impacts can compromise composite materials’ performances. The paper investigates the low-velocity impact behaviour of glass fibre-reinforced plastic laminates with various thicknesses and under different working conditions. In detail, composite laminates were subject to low-velocity impact tests by changing the temperature and the impact energy. The goal of these tests is to investigate some key features such as absorbed energy, maximum impact force, indentation depth, and delamination area. In addition, the experimental data allowed the training of artificial neural network-based models. The increase in the impact strength and the decrease in the specimens’ deflection confirmed greater rigid impact performance at low temperatures. This behaviour translates into a different onset and growth of damage, detecting lower indentation and delamination at the decrease in the temperature. Finally, the intelligent models can provide outstanding predictions, estimate complex nonlinear relationships between inputs and outputs, as well as signal the presence of anomalous data because of possible human error.